Measuring rolling shutter: put a number on this issue!

I just added the results for the BlackMagic Pocket Cinema Camera: very impressive, at 17.8ms

Even so, there are lots of complaints around about the awful rolling shutter in that camera. My guess is that this is caused by the size and weight: it is so tiny that it will move around a lot more unless you add weight or treat it very gently.
 
Just added: Sony RX10.
It's the fastest rolling shutter I've measured so far, at 14.8 ms, right in expensive-camera territory. Thanks to Wesley Byram for sending me the clip to measure this.
 
Are you guys getting radically different numbers between 24p and 60p on a cam? My G6 is giving 15.15 in 24p but only 5-6 in 60p. I've been through the calculations a dozen times and 60p is always a much lower number. Surely sensor read out has nothing to do with frame rate?
 
The measurements I took are all at 24p. In some cameras (for example my NEX-5N), it does change with frame rate, because a lower portion of the sensor is read at higher fps, and this makes read-out times faster.
 
Great work guys! Probably would help to have several independent tests of each camera. Also, any way to have one, consolidated, running list?
 
I'm not sure we would gain anything by having more than one person testing each camera: I designed the test to average out all inconsistencies and give out an objective, quantitative measure. But if you are not convinced and you have any of the cameras we have already tested, you can send me a new file and we can see if it actually changes or if it doesn't (my guess: it will fall within 2ms of the number we already have).
 
Hey, so I made some clips for you this morning with the GH4. Did it in all pertinent modes and framerates (doh, I forgot to do variable frame rate 96fps, I'll try to upload that later).
http://www.ecf.org/wp-content/uploads/2014/videos/103_PANA.zip
C4K 24p
UHD 24p, 23.97, 29.97
1080p 24p, 23.93, 29.97, 59.94

*Edit* So I should have read your directions more carfully. Is there a reason to do a 180 degree (1/50 sec) shutter? It seems that sensor readout speed shouldn't be affected by this, and less motion blur, the more accurate, right? I can reshoot again if need be.

Assuming shutter speed isn't an issue, I ran the numbers on a couple frames.
Cinema 4K 4096x2160 @ 24p I came out with .0203 seconds or 20.3 milliseconds.
1080p @ 24p I came out with .0133 seconds or about 13.3 milliseconds.

We'll see how Samuel's numbers come out.
 
Last edited:
Thanks for that! It took me a while to process, but what a nice bunch of test clips :)
The 180º shutter is just in case it matters, but it really shouldn't.

Measurements, in ms:
GH4 4K 24 --> 22,7
GH4 UHD 23 --> 22,6
GH4 UHD 24 --> 22,1
GH4 UHD 29 --> 22,7
GH4 FHD 23 --> 13,6
GH4 FHD 24 --> 13,6
GH4 FHD 29 --> 13,7
GH4 FHD 60 --> 13,7
(always average of 2 measurements, except for 4K, where I took 3 instead)

I'm summarizing this as
GH4 4K/UHD: 22.5 ms (23.2-22.3-22.7-22.7-22.4-21.8-22.4-22.8-22.6)
GH4 1080p: 13.7 ms (13.7-13.5-13.1-14.0-13.9-13.5-13.5-13.9)
 
Last edited:
Why is this in HDLR section? This isn't HDSLR issue at all as FS series have rolling shutter too. It's a CMOS sensor issue. And none of these cameras except 5D are DSLR anyway
 
Well, DSLR is an obsolete term, I think of it as "stills cameras that also shoot video". And with the exception of the BlackMagics that's what I have tested.

No other section would be better, anyway, since there's no "all cameras" section.
 
way out of my league, so, no idea
if anybody with access to them sends me a clip, I can add them to the table
 
Samuel H said:
way out of my league, so, no idea
if anybody with access to them sends me a clip, I can add them to the table
Click to expand...

But F65 is there in list. Costs more than F5 and Alexa.

By the way why is F65 there? It has global mechanical shutter.
 
I haven't ran the numbers for that second list in the bottom. The F65 is there because Squig found some numbers for it somewhere in the internet.
 
It'll be nice to see the Sony A7s results compared in its other recording modes:

1. FF in 50/60p
2. APS-C in 24/25/30p
3. APS-C in 50/60p

I would love to contribute but my unit hasn't arrived unfortunately.
Such reference would be really useful since Slashcam has revealed that the APS-C mode does better in resolving details compared to FF 50/60p mode, which of course was only observed a non moving shot.
 
You must log in or register to reply here.
Back
Top